R/Module_calcPercChangeSimple.R

In SOLV-Code/MetricsCOSEWIC: Metrics for COSEWIC status assessments.

#' calcPercChangeSimple
#'
#' this function just does a basic percent change calculation on a single vector.
#' calcPercChange() applies the same basic calc retrospectively to a data frame of Year x Stock,
#' Note: for now this simply replicates the internal function per.change.mod.fast
#' @param vec.in  a vector of values. NA are possible, but will result in NA trend (no infill for now)
#' @export
#' @examples
#' calcPercChangeSimple(vec.in = as.vector(Nile))

calcPercChangeSimple  <- function(vec.in){

  na.rm <- TRUE
  if(na.rm){vec.use <- na.omit(vec.in)} # NEED TO DISCUSS THIS
  if(!na.rm){vec.use <- vec.in}


  if(sum(is.na(vec.use)) == 0 ){
    n<-length(vec.use)
    #yrs <- 0:(n-1) # used to be 1:n, changed for consistency with MCMC calcs
    missing <- as.numeric(!is.na(vec.in)) # vector indicating which input values are missing

    ny <- 1
    yrs <- NA # index of years to use in the regression, built in the loop below
    for (i in 1:length(vec.in)) {
      if (missing[i]) {
        yrs[ny] = i
        ny = ny + 1
      }
    }
    yrs <- yrs -1 # to make years start at 0, as before

    lm.coeff <- .lm.fit(cbind(1,yrs),vec.use)$coefficients # uses model matrix that is usually created inside lm()
    #print(lm.coeff)

    pchange.raw <- ( (lm.coeff[1]+lm.coeff[2]*n) -  (lm.coeff[1]+lm.coeff[2])) / (lm.coeff[1]+lm.coeff[2]) *100
    pchange <- (exp(lm.coeff[1]+lm.coeff[2]*n) -  exp(lm.coeff[1]+lm.coeff[2])) / exp(lm.coeff[1]+lm.coeff[2]) *100

    #print(pchange)
  }

  #if(sum(is.na(vec.in)) >0 ){	pchange <- NA }


out.list <- list(pchange = pchange,intercept = lm.coeff[1],slope = lm.coeff[2] )


return(out.list)

} # end calcPercChange function